基于文本挖掘和复杂网络的事故致因重要度评估方法——以房屋市政较大以上事故为例

为充分挖掘事故调查报告中的有效信息,明确安全管理工作的内容.首先,利用文本挖掘分析事故调查报告,采用最小词频阈值文档频改进信息增益评估函数对分词结果降噪,通过回溯特征项在报告中的具体表述,提取事故致因,再构建同义词词库.然后,引入复杂网络以改进TF-IDF,综合事故致因因素的关联特征评估其重要度.最后,以房屋市政较大及...

Evaluation method of accident causes importance based on text mining and complex network:a case study of larger and above accidents in housing and municipal engineering

To fully extract the useful information from the accident investigation reports and clarify the contents of safety management work.firstly.the accident investigation reports were analyzed by using the text mining.and the minimum word frequency threshold document frequency was used to improve the information gain evaluation function so as to denoise the segmentation results.The accident causes were extracted by tracing the specific expression of characteristic items in the report.and the thesaurus of synonyms was constructed.Secondly.the complex network was introduced to improve the TF-IDF.and its importance was evaluated by synthesizing the correlation characteristics of accident cause factors.Finally.taking the larger and above work safety accidents in housing and municipal engineering as example.158 accident investigation reports from 2010 to 2019 were collected.The results showed that the lack of supervision and management was the most important factor leading to the larger and above accidents in housing and municipal engineering.This method can be used to find the causes of previous accidents.and can more comprehensively and accurately measure the importance of accident causes.